Large Interferometer For Exoplanets (LIFE): VII. Practical implementation of a five-telescope kernel-nulling beam combiner with a discussion on instrumental uncertainties and redundancy benefits

(Abridged) Context: In the previous paper in this series, we identified that a pentagonal arrangement of five telescopes, using a kernel-nulling beam combiner, shows notable advantages for some important performance metrics for a space-based mid-infrared nulling interferometer over several other considered configurations for the detection of Earth-like exoplanets around solar-type stars. Aims: We aim to produce a physical implementation of a kernel-nulling beam combiner for such a configuration, as well as a discussion of systematic and stochastic errors associated with the instrument. Methods: We developed a mathematical framework around a nulling beam combiner, and then used it along with a space interferometry simulator to identify the effects of systematic uncertainties. Results: We find that errors in the beam combiner optics, systematic phase errors and the RMS fringe tracking errors result in instrument limited performance at $\sim$4-7 $\mu$m, and zodiacal limited at $\gtrsim$10 $\mu$m. Assuming a beam splitter reflectance error of $|\Delta R| = 5\%$ and phase shift error of $\Delta\phi = 3$ degrees, we find that the fringe tracking RMS should be kept to less than 3 nm in order to be photon limited, and the systematic piston error be less than 0.5 nm to be appropriately sensitive to planets with a contrast of 1$\times 10^{-7}$ over a 4-19 $\mu$m bandpass. We also identify that the beam combiner design, with the inclusion of a well positioned shutter, provides an ability to produce robust kernel observables even if one or two collecting telescopes were to fail. The resulting four telescope combiner, when put into an X-array formation, results in a transmission map with a relative signal-to-noise ratio equivalent to 80% of the fully functioning X-array combiner.